Neural Posterior Estimation with Autoregressive Tiling for Detecting Objects in Astronomical Images

2025-10-03

AI Summary

To address the challenge of detecting and characterizing faint, overlapping astronomical sources in imaging data, this paper proposes a scalable variational inference framework based on neural posterior estimation. The method introduces a spatially autoregressive variational distribution structured via a K-color checkerboard partitioning scheme; its conditional independence structure matches that of the true posterior, substantially improving posterior calibration. Furthermore, the variational family is parameterized by a convolutional neural network (CNN) and optimized by minimizing the expectation of the forward KL divergence, enabling efficient approximation of complex, multimodal posteriors. Evaluated on Sloan Digital Sky Survey (SDSS) data, the approach achieves state-of-the-art detection performance, demonstrating high accuracy, strong robustness to source blending and noise, and computational scalability—key requirements for real-world astronomical applications.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Upcoming astronomical surveys will produce petabytes of high-resolution images of the night sky, providing information about billions of stars and galaxies. Detecting and characterizing the astronomical objects in these images is a fundamental task in astronomy -- and a challenging one, as most of these objects are faint and many visually overlap with other objects. We propose an amortized variational inference procedure to solve this instance of small-object detection. Our key innovation is a family of spatially autoregressive variational distributions that partition and order the latent space according to a $K$-color checkerboard pattern. By construction, the conditional independencies of this variational family mirror those of the posterior distribution. We fit the variational distribution, which is parameterized by a convolutional neural network, using neural posterior estimation (NPE) to minimize an expectation of the forward KL divergence. Using images from the Sloan Digital Sky Survey, our method achieves state-of-the-art performance. We further demonstrate that the proposed autoregressive structure greatly improves posterior calibration.

Problem

• Detecting faint overlapping astronomical objects in large-scale sky surveys
• Developing variational inference methods for small-object detection in images
• Improving posterior calibration for astronomical object characterization

Innovation

• Autoregressive variational distributions partition latent space
• Convolutional neural network parameterizes variational distribution
• Neural posterior estimation minimizes forward KL divergence